*BSD News Article 16461


Return to BSD News archive

Newsgroups: comp.os.386bsd.bugs
Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!saimiri.primate.wisc.edu!sdd.hp.com!crash!fpm
From: fpm@crash.cts.com (Frank Maclachlan)
Subject: [NetBSD-0.8] Problems w/ boot blocks
Organization: CTS Network Services (crash, ctsnet), El Cajon, CA
Date: 24 May 93 13:20:50 PDT
Message-ID: <1993May24.132050.29563@crash>
Lines: 28

I noticed that NetBSD-0.8 replaces the stage 2 wd boot, bootwd,
w/ Julian Elischer's boot code.  This causes a problem w/ a drive
which has more than 1024 cylinders, requires bad sector forwarding,
and has a NetBSD (or 386bsd) partition which ends beyond cylinder
1023 (mine ends at cylinder 1629).  The bad144 replacement list
and replacement sectors are at the very end of the 386bsd partition
and are not accessable by the boot code; it can't read anything
beyond cylinder 1023.  The boot code issues an error message
indicating that the bad144 table is invalid (it's reading the
wrong cylinder).  If the kernel is written over a bad sector,
that sector will not be replaced.  This is not good.

I simply replaced the bootwd w/ the original 386bsd-0.1 bootwd
and the problem went away.

Also, the bootxx binaries in NetBSD-0.8's /usr/mdec have non-zero
values in the first 512 bytes.  This will cause disklabel to
fail w/ an error something like:

	Bootstrap doesn't leave room for disk label

If this happens, simply rebuild the boot blocks.

		Regards,
		Frank MacLachlan
--
UUCP: {hplabs!hp-sdd ucsd nosc}!crash!fpm
INET: fpm@crash.cts.com